Sound Generation with Deep Learning - Approaches and Challenges

Explore the approaches and challenges of sound generation with deep learning in this comprehensive video lecture. Gain an understanding of the sound generation task and learn to classify different types of sound generation systems. Delve into various approaches and deep learning models used for generating sound, including raw audio and spectrogram-based methods. Examine the challenges encountered with different techniques and discuss the features used to train generative sound systems. Discover the advantages and limitations of various sound representations, deep learning architectures, and input types for sound generation. Conclude with insights into building a sound generative system and future directions in this exciting field of artificial intelligence and audio processing.

Intro
Defining the sound generation task
Classification of sound generation systems
Types of generated sounds
Sound representations
Generation from raw audio
Challenges of raw audio generation
Generation from spectrograms
Advantages of generation from spectrograms
Challenges of generation from spectrograms
Can we generate sound with MFCCs?
DL architectures for sound generation
Inputs for generation
Details about the sound generative system we'll build
What's next?